Entropy Mixing Networks: Enhancing Pseudo-Random Number Generators with Lightweight Dynamic Entropy Injection

Azizol Abdullah; Mohamed Aly Bouke; Mohamed Othman; Normalia Samian; Nur Izura Udzir; Omar Imhemmed Alramli; Zurina Mohd Hanapi

REVIEW 4 major objections 4 minor 21 references

Entropy Mixing Networks: Enhancing Pseudo-Random Number Generators with Lightweight Dynamic Entropy Injection

T0 review · 4 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The Entropy Mixing Network claims that a Mersenne Twister whose outputs are periodically hashed together with OS entropy outperforms SystemRandom and the plain Mersenne Twister on uniformity, entropy, and predictability, at the cost of…

desk verdict Algorithm 1's state update never actually reseeds the Mersenne Twister, so the paper's central 'entropy injection' claim and its cryptographic suitability conclusion are unsupported. read the letter →

arxiv 2501.08031 v1 pith:N5MXXNNY submitted 2025-01-14 cs.CR

Mohamed Aly Bouke , Omar Imhemmed Alramli , Azizol Abdullah , Nur Izura Udzir , Normalia Samian , Mohamed Othman , Zurina Mohd Hanapi This is my paper · ORCID

classification cs.CR

keywords EntropyMixingNetworkhybridrandomnumbergeneratorinjectionpseudo-randomcryptographicrandomnessstatisticaltestingMersenneTwisterSHA-256

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces the Entropy Mixing Network (EMN), a hybrid random number generator that periodically mixes operating-system entropy into the output of a Mersenne Twister PRNG using a SHA-256 hash and an XOR operation. It claims that this design measurably improves randomness quality over Python's SystemRandom and the plain Mersenne Twister, achieving the highest chi-squared p-value (0.9430), the highest entropy (7.9840), and the lowest predictability (-0.0286) among the three. The improvement comes with a speed cost, as EMN takes about 0.26 seconds for the benchmark run versus 0.02 seconds for the Mersenne Twister. The paper argues this trade-off makes EMN suitable for cryptographic applications where randomness quality is prioritized over throughput.

What carries the argument

The load-bearing object is the entropy-mixing step state = SHA256(state ⊕ E), executed every f cycles, followed by the output combination O = state ⊕ R. This step is what supposedly distinguishes EMN from a plain PRNG: it is a one-way, keyed-style mixing of OS entropy with the PRNG's current value before the value is exposed. The injection frequency f is the control knob that trades entropy freshness against computational cost.

What would settle it

Implement Algorithm 1 as written with the entropy source E replaced by a constant, then compare the output sequence to EMN with real os.urandom: if the two sequences are statistically indistinguishable under the paper's own metrics, the claimed entropy injection is not the cause of the reported improvement. A more direct test is to check whether the internal state of the underlying Mersenne Twister changes after an injection step; if it does not, the generator is functionally a Mersenne Twister with a filtered output, not a hybrid RNG.

Watch

Extended reading notes

Core claim

The central claim is that a deterministic pseudo-random generator can be made statistically more random by periodically folding fresh entropy into a working state and combining that state with the PRNG's output. In EMN, whenever the generation cycle reaches a multiple of the injection frequency f, 32 bytes from os.urandom are XORed with the current 256-bit state S and passed through SHA-256; the result is then XORed with the next Mersenne Twister output R to produce the final random number O. The authors report that this procedure yields better uniformity, entropy, and lag-correlation scores than both an OS-entropy-based generator (SystemRandom) and the base Mersenne Twister, and they interpret the results as evidence that secure mixing of external entropy improves randomness quality for security-critical use.

Load-bearing premise

The paper assumes that the periodic hash-mixing operation actually injects fresh entropy into the PRNG's internal state, but in the printed Algorithm 1 the mixed value is overwritten at the end of each loop by P.getrandbits(256), so the underlying Mersenne Twister state is never reseeded and only one output per cycle reflects the OS entropy.

Editorial extensions

If this is right

EMN could be deployed as a drop-in replacement for Python's random module when the application can afford the slowdown, giving closer-to-uniform output.
The evaluation framework (chi-squared, entropy, predictability, runs test, heatmaps, PSD, autocorrelation) becomes a reusable template for comparing RNGs.
If the claim holds, the cryptographic community gains a simple recipe for converting a fast PRNG into a statistically stronger generator by periodic hashed entropy mixing.
The reported trade-off quantifies that security-oriented RNGs can sacrifice an order of magnitude in speed for modest gains in statistical quality.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The pseudocode as printed suggests the mixed state S is overwritten at the end of each loop by a fresh PRNG draw, so the entropy injection touches only the current output, not the Mersenne Twister's internal state; if so, the generator is not truly hybrid and the improvement would stem from the XOR/hash step alone rather than from reseeding.
A cleaner test of the entropy-injection claim would compare EMN against a no-entropy variant where E is replaced with a fixed constant; if the statistical scores remain identical, the OS entropy is irrelevant to the claimed gains.
The reported metrics are all simple univariate tests; applying a standard cryptographic battery such as the NIST SP 800-22 suite would either strengthen or undermine the cryptographic-suitability conclusion.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear [Colleague],

Quick take: this paper's main claim does not survive a close reading of its own pseudocode. EMN is supposed to be a hybrid RNG that periodically injects OS entropy into the PRNG state, but Algorithm 1 never reseeds or modifies the Mersenne Twister state. It computes O = S xor R once per injection cycle, then overwrites S with the next MT word. So the entropy acts as a one-time mask on a single output; between injections everything is deterministic MT. That contradicts the "Entropy Injection" design principle in Section 3.1 and guts the cryptographic-suitability claim.

What the paper does do: it puts together a standard evaluation suite (chi-squared, entropy, runs, autocorrelation, PSD, heatmaps) and compares three generators. That's a reasonable template, and the visualizations are the kind of thing a student could build on. The writing is clear enough, and the idea of periodically mixing OS randomness with a PRNG is not new—Fortuna, Yarrow, and the Linux CSPRNG all do this—but the specific combination of MT plus SHA-256 plus os.urandom with a configurable frequency isn't literally in the cited literature. That's about the limit of its novelty.

Now the soft spots, in order of severity. First, the entropy-injection flaw above. It's not a statistical quibble; it means the paper's central design doesn't do what it says. If this is just a documentation bug and the implementation actually reseeds, the authors need to show code and demonstrate state changes. Second, the evaluation is a single run with no confidence intervals or significance tests. Differences like p=0.9430 vs 0.6689 are meaningless without replication. Third, the runs test numbers are internally inconsistent: observed runs around 40,000 are compared with an expected value of 79,999.50. That suggests the test was applied to a sequence half the expected length, or the expected number is wrong—either way, the table as printed doesn't support the "smallest deviation" claim.

The paper does engage with the literature, though the reference list leans on obscure or tangential citations. The bottom line: the mechanism is broken as described, the evidence is a single run, and the reported numbers have an obvious arithmetic inconsistency. This isn't ready for peer review in its current form; it needs a corrected algorithm, reproducible code, and proper statistical treatment.

My recommendation: pass on this one. If the authors fix the pseudocode and provide real reseeding with multi-run evaluations, it could be a minor contribution, but as it stands the central claim lacks support.

Regards,

[Your name]

Referee Report

4 major / 4 minor

Summary. The manuscript proposes the Entropy Mixing Network (EMN), a hybrid random number generator intended to combine a Mersenne Twister PRNG with periodic injection of OS-level entropy using SHA-256 mixing. The authors present a pseudocode implementation (Algorithm 1) and evaluate EMN against Python's SystemRandom and MersenneTwister using chi-squared, entropy, predictability, runs test, and timing metrics. They report that EMN achieves the highest chi-squared p-value (0.9430), the highest entropy (7.9840), and the lowest predictability (-0.0286), at the cost of slower generation time (0.2602 s), and conclude that EMN is particularly suitable for cryptographic applications. The paper includes an evaluation framework with visual metrics, but the experimental basis is a single run with no statistical uncertainty quantification.

Significance. If the proposed design reliably refreshed a PRNG with OS entropy and the empirical comparison were statistically sound, a lightweight hybrid generator could be of practical interest. The manuscript has some useful ingredients: a clearly stated pseudocode, a comparative evaluation across several metrics, and an attempt to combine statistical tests with visual analysis. However, the central claim is not supported: the published Algorithm 1 does not persistently inject entropy into the PRNG state, and the empirical evidence consists of single-run point estimates without error bars or significance tests. Because the main conclusion rests on these two load-bearing issues, the paper in its current form does not establish its advertised contribution.

major comments (4)

[Section 3.3, Algorithm 1] Algorithm 1 does not implement the 'Entropy Injection' design principle stated in Section 3.1. After computing the output O <- S xor R, the pseudocode executes 'Update state: S <- P.getrandbits(256)' at the end of every loop iteration. This overwrites the SHA-256-mixed value S with a fresh Mersenne Twister output before the next iteration. The Mersenne Twister object P is never reseeded or otherwise modified, and os.urandom(32) is used only to compute a single mixed value that is consumed in one output and then discarded. Consequently, the state S used in the next injection cycle is just the previous MT output, not a persistent mixed state. The construction is therefore not a hybrid generator with continuous entropy injection into the PRNG state; it is a Mersenne Twister whose outputs are occasionally XOR-masked with a hash of OS entropy for a single value. This directly undermines the central claims in the Abstract, Section 3.1, and Section 4.6 that EMN provides secure periodic entropy injection suitable for cryptographic applications.
[Section 4.5, Table 1] The runs test results are internally inconsistent. The expected number of runs is reported as 79999.50, while the observed counts are approximately 40000 (EMN: 40185, SystemRandom: 40092, MersenneTwister: 39732). A deviation of roughly 40000 from an expected value of 80000 is not a small deviation; it is approximately half the expected number of runs. The text states that 'EMN exhibits the smallest deviation from the expected value,' but even the closest observed value differs by about 39814 runs. This indicates either a misreported expected value, an incorrect sample size or test implementation, or a misunderstanding of the runs test statistic. As presented, the runs test results do not support the paper's claims about binary randomness.
[Section 4, Table 1] The empirical comparison is based on single-run point estimates with no confidence intervals, standard errors, or significance tests. Differences such as entropy 7.9840 versus 7.9822 and predictability -0.0286 versus 0.0032 are small and may well be within sampling variability; no repeated experiments over different seeds or sample sizes are reported. Similarly, the chi-squared p-value of 0.9430 versus 0.6689 is a single realization of a random variable under the null hypothesis and does not by itself establish that EMN has superior uniformity. Without an error model or repeated trials, the claim that EMN 'outperforms' SystemRandom and MersenneTwister in critical metrics is statistically unsubstantiated.
[Section 3.4 and Section 4] The evaluation framework is not specified enough to be reproducible. The manuscript does not state the sample size, the number of independent trials, the number of bins k used in Eq. (1), the value of the entropy injection frequency f in Algorithm 1, or the Mersenne Twister seed(s). The Availability section says that simulation code and parameters are available from the corresponding author upon request, but the paper itself does not provide the parameters needed to interpret Table 1. These omissions are load-bearing because the entire contribution is an empirical comparison of point estimates.

minor comments (4)

[Section 3.3, Algorithm 1] The line 'PRNG state S <- P.seed(256)' is syntactically misleading because P.seed() in Python returns None and does not produce a value to assign to S; this should be written as two separate steps.
[Section 3.4, Eq. (1)] The number of bins k used in the chi-squared test is never defined; without this value the reported chi-squared statistics and p-values in Table 1 cannot be independently verified.
[General] The manuscript refers to Figures 1, 2, and 3, but the figures themselves are not present in the provided text; the visual claims about heatmaps, PSD, PMF, and autocorrelation plots cannot be checked from this version.
[References] There are several reference formatting errors, including 'Ramanian Academy' in reference [9] (should be 'Romanian Academy') and a malformed author field in reference [11]; these should be corrected before any resubmission.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation found; the evaluation is empirical and externally benchmarked, with no fitted parameters or self-citation chains that reduce to the paper's own inputs.

full rationale

The paper's central claim is that the proposed EMN hybrid generator outperforms SystemRandom and MersenneTwister on statistical metrics. Those metrics (chi-squared, entropy, predictability, runs test, timing) are computed directly from generator outputs using standard external formulas (Eqs. 1–3), and no parameter of the generator is fitted to those same metrics after the fact. The design is not derived from the evaluation results, and no load-bearing result is justified by a self-citation or by an imported uniqueness theorem. The most serious weakness identified by the reviewer—that Algorithm 1 overwrites the mixed state with P.getrandbits(256) before the next iteration, so the Mersenne Twister is never reseeded and the entropy injection affects only one output—is a structural and security-relevant implementation flaw, not a circularity: the paper's conclusion does not reduce by construction to its input, nor does it rename a known result. Accordingly, the appropriate circularity finding is no significant circularity, score 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim depends on unstated tuning parameters (injection frequency, sample size) and standard cryptographic assumptions about SHA-256 and os.urandom. No new physical or mathematical entities are introduced.

free parameters (3)

entropy injection frequency f
Configurable parameter in Algorithm 1; its value is never specified, yet it determines how often OS entropy enters the output and therefore affects all randomness metrics.
sample size for statistical tests
The number of generated values and the bit length used in chi-squared, entropy, runs, and correlation tests are not stated, making the reported statistics unverifiable.
chi-squared bin count k = 256
Implied by the stated entropy maximum of 8 bits and used to compute the p-values; the choice is not justified but is conventional for byte-oriented outputs.

assumptions (4)

domain assumption SHA-256 behaves as a pseudorandom function and produces uniform, unpredictable output when mixed with entropy.
The secure mixing step in Section 3.1 assumes this property without proof or reference to a security model.
domain assumption os.urandom() provides high-quality unpredictable entropy from the operating system.
The entropy capture in Algorithm 1 relies on OS randomness being a valid source of true entropy; standard but asserted without discussion.
domain assumption A higher chi-squared p-value indicates better randomness and can be used to rank generators.
Section 4.1 interprets the highest p-value as the best uniformity, but all p-values exceed typical thresholds, and ranking by p-value is not statistically meaningful.
domain assumption A single run of each statistical test is representative of generator quality.
The results table reports one realization with no repetitions or confidence intervals, implicitly assuming stability across seeds and sample sizes.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Entropy Mixing Networks: Enhancing Pseudo-Random Number Generators with Lightweight Dynamic Entropy Injection." pith.science (2026). https://pith.science/paper/N5MXXNNY

@misc{pith2026250108031,
  author       = {Pith},
  title        = {Pith review of: Entropy Mixing Networks: Enhancing Pseudo-Random Number Generators with Lightweight Dynamic Entropy Injection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N5MXXNNY}},
  note         = {Machine review of arXiv:2501.08031}
}

read the original abstract

Random number generation plays a vital role in cryptographic systems and computational applications, where uniformity, unpredictability, and robustness are essential. This paper presents the Entropy Mixing Network (EMN), a novel hybrid random number generator designed to enhance randomness quality by combining deterministic pseudo-random generation with periodic entropy injection. To evaluate its effectiveness, we propose a comprehensive assessment framework that integrates statistical tests, advanced metrics, and visual analyses, providing a holistic view of randomness quality, predictability, and computational efficiency. The results demonstrate that EMN outperforms Python's SystemRandom and MersenneTwister in critical metrics, achieving the highest Chi-squared p-value (0.9430), entropy (7.9840), and lowest predictability (-0.0286). These improvements come with a trade-off in computational performance, as EMN incurs a higher generation time (0.2602 seconds). Despite this, its superior randomness quality makes it particularly suitable for cryptographic applications where security is prioritized over speed.

Figures

Figures reproduced from arXiv: 2501.08031 by the authors.

**Figure 1.** Visual Metrics for EMN: (a) Heatmap of Correlation, (b) Power Spectrum Density, (c) Probability Mass Function, (d) Autocorrelation Function. 4.6 Implications and Trade-offs The results underscore the trade-offs between randomness, quality, security, and performance: • EMN: Offers the best randomness quality and resistance to predictability at the cost of slower high-frequency performance. This makes it ideal for c… view at source ↗

**Figure 2.** Visual Metrics for SystemRandom: (a) Heatmap of Correlation, (b) Power Spectrum Density, (c) Probability Mass Function, (d) Autocorrelation Function [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 19 canonical work pages

[1]

Memristive technologies for data storage, computation, encryption, and radio-frequency communication

Mario Lanza et al. “Memristive technologies for data storage, computation, encryption, and radio-frequency communication”. In: Science 376.6597 (2022), eabj9979

work page 2022
[2]

Entropy sources based on silicon chips: True random number genera- tor and physical unclonable function

Yuan Cao et al. “Entropy sources based on silicon chips: True random number genera- tor and physical unclonable function”. In: En- tropy 24.11 (2022), p. 1566

work page 2022
[3]

Vera: Vector-based ran- dom matrix adaptation

Dawid J Kopiczko, Tijmen Blankevoort, and Yuki M Asano. “Vera: Vector-based ran- dom matrix adaptation”. In: arXiv preprint arXiv:2310.11454 (2023)

arXiv 2023
[4]

Physical Security in the Post-quantum Era: A Survey on Side-channel Analysis, Random Number Generators, and Physically Unclonable Functions

Sreeja Chowdhury et al. “Physical security in the post-quantum era: A survey on side- channel analysis, random number genera- tors, and physically unclonable functions”. In: arXiv preprint arXiv:2005.04344 (2020)

work page Pith review arXiv 2020
[5]

Experimental quan- tum key distribution certified by Bell’s theo- rem

David P Nadlinger et al. “Experimental quan- tum key distribution certified by Bell’s theo- rem”. In: Nature 607.7920 (2022), pp. 682– 686

work page 2022
[6]

A chaos-metastability TRNG for natively flexible IGZO circuits

Krzysztof Gołofit, Piotr Z Wieczorek, and Marcin Pilarz. “A chaos-metastability TRNG for natively flexible IGZO circuits”. In: AEU- International Journal of Electronics and Com- munications 170 (2023), p. 154835

work page 2023
[7]

Design and analysis of a true random number generator based on GSR signals for body sensor networks

Carmen Camara et al. “Design and analysis of a true random number generator based on GSR signals for body sensor networks”. In: Sensors 19.9 (2019), p. 2033

work page 2019
[8]

Secure blind watermarking using Fractional-Order Lorenz system in the frequency domain

Sherif H AbdElHaleem, Salwa K Abd-El- Hafiz, and Ahmed G Radwan. “Secure blind watermarking using Fractional-Order Lorenz system in the frequency domain”. In: AEU- International Journal of Electronics and Com- munications 173 (2024), p. 154998. Published under Creative Commons Attribution license. ArXiv.org e-Print archive — DOI: https://doi.org/xxxxxxx 9/9

work page 2024

Show all 21 references

[9]

Generation and testing of random numbers for cryptographic appli- cations

Kinga Marton et al. “Generation and testing of random numbers for cryptographic appli- cations”. In: Proceedings of the Ramanian Academy, Series A 13.4 (2012), pp. 368–377

2012
[10]

Random Number Generators: Principles and Applications

Anastasios Bikos et al. “Random Number Generators: Principles and Applications”. In: Cryptography 7.4 (2023), p. 54

2023
[11]

A statistical test suite for random and pseudorandom number generators for cryptographic applications

Elaine Barker Smid et al. “A statistical test suite for random and pseudorandom number generators for cryptographic applications”. In: Her research interest includes Computer se- curity, secure operating systems, Access con- trol, Distributed systems, Intrusion detection syste...

2010
[12]

Memristive true random number generator for security applications

Xianyue Zhao et al. “Memristive true random number generator for security applications”. In: Sensors 24.15 (2024), p. 5001

2024
[13]

A comprehensive review of quantum random number generators: Con- cepts, classification and the origin of random- ness

Vaisakh Mannalatha, Sandeep Mishra, and Anirban Pathak. “A comprehensive review of quantum random number generators: Con- cepts, classification and the origin of random- ness”. In: Quantum Information Processing 22.12 (2023), p. 439

2023
[14]

True random number generators

Mario Stip ˇcevi´c and C ¸ etin Kaya Koc ¸. “True random number generators”. In: Open prob- lems in mathematics and computational sci- ence. Springer, 2014, pp. 275–315

2014
[15]

A search for good pseudo-random num- ber generators: Survey and empirical stud- ies

Kamalika Bhattacharjee and Sukanta Das. “A search for good pseudo-random num- ber generators: Survey and empirical stud- ies”. In: Computer Science Review 45 (2022), p. 100471

2022
[16]

A guideline on pseudo- random number generation (PRNG) in the IoT

Peter Kietzmann, Thomas C Schmidt, and Matthias W ¨ahlisch. “A guideline on pseudo- random number generation (PRNG) in the IoT”. In: ACM Computing Surveys (CSUR) 54.6 (2021), pp. 1–38

2021
[17]

A new method for hy- brid pseudo random number generator

Erdinc ¸ Avaroglu et al. “A new method for hy- brid pseudo random number generator”. In: Informacije MIDEM 44.4 (2014), pp. 303– 311

2014
[18]

Hybrid pseudo- random number generator for cryptographic systems

Erdinc ¸ Avaro ˘glu et al. “Hybrid pseudo- random number generator for cryptographic systems”. In: Nonlinear Dynamics 82 (2015), pp. 239–248

2015
[19]

Random Number Genera- tors and Seeding for Differential Privacy

Naoise Holohan. “Random Number Genera- tors and Seeding for Differential Privacy”. In: arXiv preprint arXiv:2307.03543 (2023)

2023 arXiv
[20]

Mersenne twister: a 623-dimensionally equidistributed uniform pseudo-random number generator

Makoto Matsumoto and Takuji Nishimura. “Mersenne twister: a 623-dimensionally equidistributed uniform pseudo-random number generator”. In: ACM Transactions on Modeling and Computer Simulation (TOMACS) 8.1 (1998), pp. 3–30

1998
[21]

A new pseudo-random number generator

A Rotenberg. “A new pseudo-random number generator”. In: Journal of the ACM (JACM) 7.1 (1960), pp. 75–77. Published under Creative Commons Attribution license

1960

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

Memristive technologies for data storage, computation, encryption, and radio-frequency communication

Mario Lanza et al. “Memristive technologies for data storage, computation, encryption, and radio-frequency communication”. In: Science 376.6597 (2022), eabj9979

work page 2022

[2] [2]

Entropy sources based on silicon chips: True random number genera- tor and physical unclonable function

Yuan Cao et al. “Entropy sources based on silicon chips: True random number genera- tor and physical unclonable function”. In: En- tropy 24.11 (2022), p. 1566

work page 2022

[3] [3]

Vera: Vector-based ran- dom matrix adaptation

Dawid J Kopiczko, Tijmen Blankevoort, and Yuki M Asano. “Vera: Vector-based ran- dom matrix adaptation”. In: arXiv preprint arXiv:2310.11454 (2023)

arXiv 2023

[4] [4]

Physical Security in the Post-quantum Era: A Survey on Side-channel Analysis, Random Number Generators, and Physically Unclonable Functions

Sreeja Chowdhury et al. “Physical security in the post-quantum era: A survey on side- channel analysis, random number genera- tors, and physically unclonable functions”. In: arXiv preprint arXiv:2005.04344 (2020)

work page Pith review arXiv 2020

[5] [5]

Experimental quan- tum key distribution certified by Bell’s theo- rem

David P Nadlinger et al. “Experimental quan- tum key distribution certified by Bell’s theo- rem”. In: Nature 607.7920 (2022), pp. 682– 686

work page 2022

[6] [6]

A chaos-metastability TRNG for natively flexible IGZO circuits

Krzysztof Gołofit, Piotr Z Wieczorek, and Marcin Pilarz. “A chaos-metastability TRNG for natively flexible IGZO circuits”. In: AEU- International Journal of Electronics and Com- munications 170 (2023), p. 154835

work page 2023

[7] [7]

Design and analysis of a true random number generator based on GSR signals for body sensor networks

Carmen Camara et al. “Design and analysis of a true random number generator based on GSR signals for body sensor networks”. In: Sensors 19.9 (2019), p. 2033

work page 2019

[8] [8]

Secure blind watermarking using Fractional-Order Lorenz system in the frequency domain

Sherif H AbdElHaleem, Salwa K Abd-El- Hafiz, and Ahmed G Radwan. “Secure blind watermarking using Fractional-Order Lorenz system in the frequency domain”. In: AEU- International Journal of Electronics and Com- munications 173 (2024), p. 154998. Published under Creative Commons Attribution license. ArXiv.org e-Print archive — DOI: https://doi.org/xxxxxxx 9/9

work page 2024